GitHub integration and incremental refresh – how to avoid overwriting data?

[ua77]

I have a Power BI semantic model with incremental refresh enabled. I am using a GitHub connection for version control, and I make changes in a dev workspace connected to a new branch before merging to main. (that is connected to prod workspace). Currently, when I merge to main, data is overwritten.

How can I ensure that existing data in historical partitions remains intact when merging changes of the semantic model to production?

[anilelmastasi]

Hello @ua77 , with overwriting what did you mean? Partitions disappeared? Full refresh triggered? Or actual historical data is lost?

[ua77]

Hi,
I am experiencing both partitions that have disappeared and historical data that is lost.

When I accept incoming changes from main, all existing partitions (and their data) disappear. When I trigger a refresh of the semantic model, it starts a full refresh and rebuilds all partitions from scratch. The problem is that the full refresh takes a long time and causes downtime for the reports (until the refresh is completed).